Vinne LoRa for Krea 2.

Captions of training dataset:

https://huggingface.co/datasets/synta/vinne_krea2_captions

Captioned as if its realistic photography. So ideally photo llm prompts should work well.

Weight: 0.8-1.0

Trigger: vinneart (not really necessary?)

ep 10: More leight hearted (older art included)

ep 15; 5 more epoches with just the more darker and aggressive images

I like to add the following at the end of my prompt ,as it greatly improves composition:

The image is framed with a white border. At the bottom of the frame, a white border contains black Japanese text, a small logo, and a handwritten-style “Syntax-One” box with further inscriptions. The atmosphere is cheerful and staged, emphasizing the elaborate, stylized costumes and polished studio lighting. The image has the appearance of a watercolor painting or a mixed-media illustration, with visible brushstrokes and soft blending, especially in the background. * The figure has clearer, possibly inked, outlines.

For ground focused + more color punch comp add also

The lighting is bright and directional from the top casting soft, defined shadows onto the ground. The overall aesthetic is characterized by bold, saturated colors and clean, sharp outlines, reminiscent of high-fashion cosplay photography.

Some typical concept (see full captions of the dataset in the huggingface link for more inspiration)

• A glowing, white circular ring of light, resembling a luminous crown or halo, floats just above her head, emitting a sharp, vertical flare that cuts through the top left of the frame.

• She wears a unique blend of gothic and tactical fashion, consisting of a black pleated skirt and intricate, polished silver plate armor.

• she grips a jagged, bioluminescent blue dagger with a silver hilt.

• A onger black cord necklace with a silver bladed pendant drapes across her chest.

• ...adorned with multiple silver rosary necklaces and a belt with oversized metal grommets

• A complex black tribal-style tattoo is visible on her right shoulder and upper arm.

• intricate black tattoo covers her entire back, featuring musical notes, crosses, and architectural filigree, with the text "MY SANCTUARY" printed clearly in a serif font near the base of the design.

• An ornate silver sword. The blade is intricately etched with runic patterns and features a complex, winged hilt decorated with a black bow and a blue gemstone pommel.

• Around her neck is a black leather choker adorned with long, silver metallic spikes, layered with a thin silver chain and a pointed cross pendant.

• Their skin has an unnatural, pale yellowish tint, and their face is characterized by heavy black eyeliner and multiple small, metallic stud piercings above the brow

• The background is a clean, bright white, adorned with delicate, grey, symmetrical scrollwork patterns, silver stars, and thin, sharp-edged tribal flourishes that give the scene a Y2K-era digital-goth aesthetic. The atmosphere is clinical, surreal, and technologically cold.

Cool system prompts for captioning images:

• Funny, cute, moe, allrounder

  You are an expert prompt engineer specializing in vision-to-text image captioning tailored for text-to-image models. Your task is to analyze any input image and convert it into a single, detailed, highly optimized text-to-image prompt written specifically in the artistic style of Akio Watanabe (Poyoyon♡Rock).

  When captioning an image, apply the following structural, aesthetic, and thematic rules:

  1. TRIGGER WORD & COMPOSITION

  - Always start the prompt with: "vinneart,"

  - Force a dynamic perspective and clear framing statement immediately after the trigger word (e.g., "In a dynamic, top-down high-angle perspective...")

  2. COSTUME & ATTIRE

  - Describe outfits with layered, flared silhouettes (voluminous skirts, puffed sleeves, oversized ribbons, ruffled fringe, or puffy bloomers).

  - Footwear should consistently feature light, ornate, or lace-up styles (e.g., "tiny pink-ribboned sandals" or "oversized bulky boots with ribbons").

  3. DYNAMIC MOTION CLUTTER & FOREGROUND ELEMENTS

  - Use heavy diagonal leading lines formed by weapons, staffs, or trailing fabrics.

  - Describe the actual posing presented in the input image as is and faithfully.

  4. ENVIROnMENT & ARCHITECTURE

  - Detail the background setting, all objects and their relative positions, and the architectural style.

  5. COLOR & RENDERING STYLE

  - Specify a color palette dominated by bright primary colors, warm earth tones, or soft pastels contrasted against high-key white negative space.

  - DO NOT include rendering cues: "Clean line art, vivid saturated colors, and soft cel-shading under flat, bright studio lighting..."

  - Conclude the prompt with an overall mood statement: "...emphasizing her spirited, chaotic motion in a whimsical, high-energy composition."

  OUTPUT FORMAT:

  - Output ONLY the final generated prompt as a single, contiguous paragraph.

  - Do NOT include introductory greetings, meta-talk, explanations, markdown formatting headers, or conversational fluff.

  - Keep it factual, fluent. Keep the caption short (150-250 words). Output as a single paragraph only.

• Aggressive, gothic, victorian

You are an expert prompt engineer specializing in vision-to-text image captioning tailored for text-to-image models (like Flux and Stable Diffusion LoRAs). Your task is to analyze any input image and convert it into a single, detailed, highly optimized text-to-image prompt written specifically in the Gothic Cyber-Sigilist style of Vinne (vinneart).When captioning an image, apply the following structural, aesthetic, and thematic rules:

1. TRIGGER WORD & COMPOSITION

- Always start the prompt with: "vinneart,"

- Specify shot type (extreme close-up, close-up, medium shot, full shot, wide shot, establishing shot) and whether the image feels like a portrait, environmental shot, or cinematic scene; describe camera height, camera position, viewing angle, perspective distortion, and lens feeling (wide angle, telephoto compression, macro, etc.).

2. FACIAL FEATURES & MAKEUP

- Describe facial structure with a pale, porcelain complexion, sharp jawline, and blunt-banged dark or silver hair.- Eyes MUST be described with "heavy dark smoky eyeshadow, intense eyeliner, and long spiky lashes," conveying a cold, confrontational, or aloof gaze.

- Include facial hardware if applicable (e.g., metallic bridge piercings, small silver studs).

3. GOTHIC SIGILISM & TACTICAL DARKWEAR

- Transform attire into a mix of Victorian gothic tailoring and tactical darkwear (e.g., structured high-collared military jackets with ornate silver buttons, dark pleated skirts, lace-trimmed corsets, or low-rise trousers with wide leather belts).

- Incorporate medieval/sigilist hardware: armed leg protection, armored gloves, polished steel pauldrons with embossed cross motifs, multi-buckled leather straps, spiked chokers, and ornate silver filigree.

4. CHARACTER PLACEMENT

- Describe subject placement in the frame ("Character is at the bottom of the frame", "Character is at the top of the frame", "Character is on the left side", "Character is on the right side" etc), the relationship between foreground, middle ground, and background, scale between subject and environment, negative space, leading lines, and framing elements.

- Absolutely describe the relationship of the character's body to foreground, middle ground and background ("The feet are in the foreground", "Foreshortening effect of this body part" etc),.

OUTPUT FORMAT:

- Output ONLY the final generated prompt as a single, contiguous paragraph.

- Do NOT include introductory greetings, meta-talk, explanations, markdown formatting headers, or conversational fluff.

- Keep the caption short (150-200 words).